document() error handling

The spec doesn't say what should happen if the URL constructed by document()
is an invalid URL, or is a valid URL pointing to no resource, or is a URL
pointing to a resource that cannot be retrieved.

Possible actions are:
- omit the document silently from the returned nodeset
- output a message (somewhere?) and omit the document from the returned
nodeset
- abandon processing

One feels this ought to be under user control!

Mike Kay